National grain reserve is important in terms of responding to disasters and the unbalance between supply and demand in many countries. In China, the government supplements grain supply through online auctions. This study focuses on the auction policy of national grain reserve. We develop an agent-based simulation model of China’s wheat market with detail descriptions of different agents, including national grain reserve, grain trading enterprises and grain processing enterprises. Based on this model, the Optimal Computing Budget Allocation (OCBA) simulation optimization method is adopted to analyze the characteristics of optimal decision variables under different scenarios, with an objective to minimize the fluctuation of wheat price. We obtain some insights about operations of national grain reserve. As the first agent-based simulation model about national grain reserve and grain market, this model can be widely used in agricultural economics, and can provide policy supports to the government.
Grain security is one of the most important issues worldwide. The stable development of grain market is the footstone of a country’s economy system. Since China is a country feeding about 19% of the world’s population using only 7% of arable land, a stable grain market price is critical to ensure social stability. This work is motivated by China’s national grain reserve auction policy, and we establish an agent-based simulation model to characterize grain market participants’ behavior to find some insights to support the government’s grain auction strategy.
In China, the government sets a minimum purchase price every year. When grain market price is lower than that price, the government starts to procure grain until market price is higher than that price. To supplement grain supply, the government releases grain to the market through an online auction mechanism. In accordance with the mechanism, national grain reserve (NGR) decides the minimum sale price, the auction quantity and the auction time interval. The government will hold an auction periodically and public the minimum sale price and the auction quantity to all bidders. When getting auction message, bidders can bid for auction items with a bidding price higher than the minimum sale price, and bidders with higher price will have priority in bidding the auction. According to historical auction data of wheat and rice during last three years, an auction time interval is one week, and the auction quantities stay the same. Under this auction strategy, the average successful transaction rate is less than 25%. While a lot of public resources are occupied by NGR, the effect of stabilizing market price is unnoticeable. To solve this issue, we first establish an agent-based simulation model to describe the auction mechanism and grain market participants’ decision strategy. Then, we evaluate NGR’s optimal decision variables (includes the auction time interval, the auction quantity and the minimum sale price) to minimize the fluctuation of grain market price.
Existing literatures on grain market policy can be classified into three research issues: price policies, production supporting policies and grain reserve policies. Many researches use computable general equilibrium models to analyze the effect of price policies on grain market, such as Hosoe (2004), Yu and Jensen (2010) and Mason and Myers (2013). Besides, some researches focus on evaluating production supporting policies’ regulation effects to grain market (Hansen, Tuan, and Somwaru 2011). However, there are few research studies about grain reserve policy, and this paper is motivated by the issue. Miao and Zhong (2006) analyzes the correlations between grain reserve and market price using the annual data from 1978 to 2005, and indicates that national grain reserve affects grain market price from its incremental level and stock level. Wang and Li (2013) analyzes the effects of state-owned grain auction on market price using vector autoregressive approach, and concludes that auction performs as a signal of demand and supply rather than a balance tool under the scenario of abundant supplies. These researches analyze the effects of national grain reserve from the view of macro-economy, and ignore the behavior strategy of different market agents and their interactions, which are quite important to evaluate the policy influences.
To mirror the real grain reserve system as closely as possible, in this paper, we develop a simulation model that comprises all participants of the grain reserve system, and simulate their operations and interactions. Agent-based modeling (ABM) offers such a conceptual framework to solve this modeling problem, because it simulates market participants as autonomous decision-making and heterogeneous entities called agents. Each agent individually assesses its decision environment, and makes decisions based on a set of rules. Besides, it is flexible and it can capture emergent phenomena (Bonabeau 2002). Although there is no direct study about agent-based modeling of grain reserve system, this method has been applied to evaluate policy effect in other fields, and obtains some influential models. Happe, Balmann, and Kellermann (2004) builds the Agricultural Policy Simulator (AgriPoliS), it analyzes farmers’ structural change and endogenous adjustment reactions in response to a policy change, and many researches analyze effects of different policies or in different regions based on this model (Bert et al. 2011; Zimmermann and Heckelei 2012). In the electricity market, an Electricity Market Complex Adaptive Systems (EMCAS) simulates every participant’s behavior, and it is used to evaluate efficiencies of different regulation policies (North et al. ). A lot of researchers analyze different countries’ regulation policies based on EMCAS model (North et al. ; Conzelmann et al. ). All above literatures prove the capability of agent-based model in policy analysis field, which motivates us to develop such a simulation platform to analyze the effectiveness of grain reserve policy.
In this paper, we develop an agent-based simulation model of national grain market with detail descriptions of different agents including state-owned national grain reserve (NGR), the wholesaler grain trading enterprises (GTE), and grain processing enterprises (GPE). Based on this model, we analyze the optimal decision variables of grain reserve auction policy ( includes the auction time interval,the auction quantity and the minimum sale price) under different scenarios using optimal computing budget allocation (OCBA) method, aiming at minimizing price influences. Some insights of NGR’s operation strategy under current auction policy are provided in this research.